Redox Properties of Manganese-Containing Zirconia Solid Solution

May 27, 2011 - Department of Inorganic Chemistry, Fritz Haber Institute of the Max Planck .... O. S. Reu , A. V. Palii , S. M. Ostrovsky , P. L. W. Tr...
0 downloads 0 Views 2MB Size
ARTICLE pubs.acs.org/JPCA

Redox Properties of Manganese-Containing Zirconia Solid Solution Catalysts Analyzed by In Situ UVVis Spectroscopy and Crystal Field Theory Sophia I. Klokishner,*,† Oleg Reu,† Carine E. Chan-Thaw, Friederike C. Jentoft,*,‡ and Robert Schl€ogl Department of Inorganic Chemistry, Fritz Haber Institute of the Max Planck Society, Faradayweg 4-6, 14195 Berlin, Germany ABSTRACT: The optical absorption spectra of manganesepromoted sulfated zirconia, a highly active alkane isomerization catalyst, were found to be characterized by oxygen-to-manganese charge-transfer transitions at 300320 nm and dd transitions of manganese ions at 580 and 680 nm. The latter were attributed to Mn4þ and Mn3þ ions, which are known to be incorporated in the zirconia lattice. The oxygen surroundings of these ions were modeled assuming a substitutional solid solution. The crystal field splittings, vibronic coupling constants, and oscillator strengths of the manganese ions were calculated on the basis of a cluster model that considers the manganese center as a complex with the adjacent ions of the lattice as ligands. The ratio of Mn3þ to Mn4þ ions was determined using the spectra and the model, and the relative concentrations of Mn2þ, Mn3þ, and Mn4þ ions were determined with the help of the average valence known from X-ray absorption data in the literature. The redox behavior of manganese-promoted sulfated zirconia in oxidizing and inert atmosphere was elucidated at temperatures ranging from 323 to 773 K.

1. INTRODUCTION Manganese is a component in a variety of solid catalysts. Mixed coppermanganese oxides serve as CO oxidation catalysts in breathing apparatuses,1,2 and MnO2 is used as a catalyst for the hydration of nitriles to amides.3 The addition of manganese and copper to titania results in catalysts for oxidation of ammonia to N2,4 and addition of manganese to zirconia produces materials active in the total oxidation of hydrocarbons.5,6 Manganese also functions as a promoter in sulfated zirconia catalysts, which are highly active for the skeletal isomerization of n-butane at low temperatures.7,8 Different manganese oxidation states are stable, and mixedvalence manganese oxides such as Mn3O4 (hausmannite) exist. Thus, manganese in catalysts (and other materials) can assume various and also multiple valences. The valence can change with activation of the catalyst and with exposure to reaction conditions. In fact, in oxidation catalysis, the ability of the active metal species to change valence is essential. The catalytic cycle then consists of reduction and reoxidation steps, and a mixture of valences can be encountered in an operating oxidation catalyst. Skeletal isomerization of alkanes is typified as an acid-catalyzed reaction, and when the promoting effect of manganese on sulfated zirconia materials was discovered, no attention was paid to the valence of manganese. The increase in activity7 was ascribed to stronger acidity of the material.7,9 However, Wan et al.10 and later also Song and Kydd11 found ironmanganesepromoted sulfated zirconia catalysts to become more active through an oxidizing treatment and proposed that n-butane is activated through oxidation (rather than acid attack), with the promoters being the redox-active sites. r 2011 American Chemical Society

If manganese activated n-butane through oxidation, it would have to be reduced during the isomerization reaction, but Jentoft et al.12 found that reduction during operation did not always occur. Klose et al.13 demonstrated that the presence of manganese or iron increases the oxidation potential of the sulfate, which had previously been shown to activate butane in an oxidative dehydrogenation step.14,15 This result indeed suggests a more indirect role of manganese but leaves unanswered the question of its most active oxidation state as a promoter of sulfated zirconia. The maximum isomerization activity of manganese-promoted sulfated zirconia was found to increase with increasing average manganese valence after activation.12 The valence is typically a noninteger number between 2 and 3;12 thus, multiple manganese species must be present in manganese-containing zirconia, as also detected by Valigi et al.16 Information on the manganese valence can be obtained by several means, for example, through dissolution of the sample and titration; however, digestion is not always possible without altering the oxidation state and cannot be applied in situ. The manganese oxidation state can be determined by X-ray absorption spectroscopy. The position of the Mn K edge was shown to depend linearly on the average manganese oxidation state,17 and this method was applied by Jentoft et al.;12 however, the fractions of the individual oxidation states in a mixed-valence sample cannot be extracted. Electron paramagnetic resonance is a suitable method to identify some of the paramagnetic states of Received: April 12, 2011 Revised: May 25, 2011 Published: May 27, 2011 8100

dx.doi.org/10.1021/jp2034054 | J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Table 1. Reported Assignments of Electronic Transitions of Manganese Species wavelength (nm)

transition (CT or dd) 2

f Mn



material

255

O

313 320

6 A1 f 4T1 Mn2þ 2 3þ O f Mn (Oh)

340

Mn2þ

6

363

Mn



6

370

O2 f Mn3þ (Oh)

370

O2 f Mn4þ or dd of Mn4þ Mn/ZrO2

404

Mn2þ

6

420

Mn2þ (cubic)

6

422 440

Mn2þ Mn2þ

6

444500

Mn3þ

5



4

469

Mn4þ

476

Mn



480

Mn2þ (cubic)

Mn

(Td)



485

Mn

485

Mn3þ (Oh)

A1 f 4E A1 f 4T2

Mn3O4

24

Mn(Ac)2 3 4H2O Mn3O4

26 24

2. EXPERIMENTAL SECTION

Mn(Ac)2 3 4H2O

26

Mn(Ac)2 3 4H2O

26

Mn2O3

24

2.1. Sample Preparation. Manganese-promoted sulfated zirconia (MnSZ) with a manganese content of 0.5 or 2.0 wt % (metal) was prepared according to the incipient wetness method as previously described.37 Sulfated hydrous zirconia (XZO 682/ 01, MEL Chemicals) was dried at 383 K for 21 h and cooled in a desiccator. A solution of Mn(NO3)2 (Merck) was added dropwise to the powder under vigorous stirring. After being allowed to dry at room temperature, the raw material was calcined in portions of 2025 g in a quartz boat, which was placed in a 29mm-i.d. quartz tube. The tube was purged with 200 mL min1 synthetic air and heated in an oven at 3 K min1 to 923 K, held at this temperature for 3 h, and then cooled. 2.2. UVVisibleNear-Infrared (UVvisNIR) Spectroscopy. A Perkin-Elmer Lambda 950 spectrometer equipped with a Harrick Praying Mantis DRP-P72 diffuse reflectance accessory and an HVC-VUV reaction chamber was used to record UVvisNIR spectra of MnSZ during various treatments. The spectroscopic measurements were carried out in the range of 2502500 nm with a scan speed of 225 nm min1. The slit width was 2.5 nm in the UVvis range and 2 nm in the NIR range, the response time was 0.2 s, and the reference beam was attenuated to 10%. The reflectance F(λ) of the sample was measured relative to a white standard, Spectralon (Labsphere). A background correction (“Autozero”) was performed with a small disk of this standard in the reaction chamber. Approximately 200 mg of catalyst was loaded into the reaction chamber. The powder was placed on a wire mesh, and the gases were streamed through this fixed bed in down-flow operation. Pure helium or O2 (both 99.999% purity from Westfalen) was used for treatments in inert or oxidizing atmosphere, at a flow of 20 mL min1 (He) or 40 mL min1 (O2). A spectrum was recorded at 323 K, and then the temperature was raised to 773 K (He) or 823 K (O2) at a rate of 5 K min1 while spectra were continuously acquired. To present and analyze the spectra, the KubelkaMunk function F(F) = (1 F)2/(2F) was calculated from the reflectance F.

A1 f 4A1, 4E Mn(Ac)2 3 4H2O A1 f 4A1, 4E MnO

A1 f 4A1 6 A1 f 4T2

ref

23 26 22

Mn/Al2O3 Mn(Ac)2 3 4H2O

25 26

Mn/Al2O3

25

4

Mn/Al2O3

27

a

CsMn(SO4) 3 12H2O 28

Eg f 5T2g A2 f 4T2 A2 f 4T2 A1 f 4F2

6

MnO

22

a

Mn/Al2O3

29

Mn2O3

24

B1g f 5B2g B1g f 5Eg

5 5

490

Mn3þ (Oh)

Mn-APO-34

30

500

Mn3þ

[Mn(H2O)6]3þ

31

500

Mn3þ of Mn2O3

Mn/ZrO2

23

535

Mn3þ

a

Mn/Al2O3

29

541

Mn2þ

6

565

Mn2þ (Td or Oh) 6A1 f 4T1

610 625

Mn2þ (cubic) Mn4þ

740

Mn3þ of Mn2O3

755

Mn3þ (Oh)

900

Mn3þ of Mn2O3

1750 a

Mn3þ (Oh)

A1 f 4T1 A1 f 4F1 A1 f 4Tl

6 4

B1g f 5A1g

5

E f 5T2

5

zirconia.16,3336 Earlier work showed that manganese-promoted sulfated zirconia catalysts contain such solid solutions,36 and we assume that the manganese ions are in a defined geometry in the zirconia lattice.

Mn(Ac)2 3 4H2O

26

Mn2þ/TiO2

24

MnO Mn/TiO2

22 32

Mn/ZrO2

23

Mn2O3

24

Mn/ZrO2

23

Mn3O4

24

See text for transition.

manganese such as Mn2þ,18 but conventional electron paramagnetic resonance (EPR) spectroscopy fails with low-spin complexes of Mn3þ 19 and sometimes even with Mn4þ complexes. X-ray photoelectron spectroscopy is another possibility to identify various manganese oxidation states if a high enough concentration of surface species is present. Unfortunately, the position of the Mn 2p signal depends on the valence and on the coordination of the cation, and some oxidation states might not be distinguishable.4,20,21 UVvisible spectroscopy has also been applied to identify the manganese valence in pure and mixed manganese oxides2224 and in supported moieties,25 but many transitions of various manganese cations fall into the same spectral region, and assignments presented in the literature are not unambiguous (Table 1). The goals of this work were to establish a consistent assignment of the transitions in the optical spectra of manganesecontaining materials with the help of a structural model of manganese ions in zirconia and calculation of their electronic transitions and to apply these assignments to the interpretation of spectra of manganese-promoted sulfated zirconia during treatment in various conditions. A key for the development of the model is the fact that manganese forms solid solutions with

3. RESULTS AND QUALITATIVE INTERPRETATION OF SPECTRA Spectra were recorded during a typical activation procedure as it would be applied prior to isomerization catalysis. MnSZ (2 wt % Mn) was heated in helium flow from 323 to 773 K and then held at 773 K for 30 min. The apparent absorption spectra of this material, obtained by converting the reflectance into the Kubelka Munk function, are shown in Figure 1. Interpretation of these spectra is facilitated by the knowledge that the average valence of manganese in the calcined catalysts is about 2.70 ( 0.05.12 Moreover, Zr4þ is not easily reduced, and no significant contributions from Zr3þ dd transitions are expected. The band gap of ZrO2 has been reported to be between 250 and 285 nm3840 and accounts for the increasing absorbance below 300 nm. Absorptions of sulfur(VI)oxygen surface species can be excluded; electronic transitions in SO3 and H2SO4 occur at wavelengths shorter than 200 nm.41 The three observed bands 8101

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

Figure 1. Diffuse reflectance UVvis spectra of manganese-promoted sulfated zirconia (2.0 wt % Mn) in a 20 mL/min helium flow in the temperature range of 323773 K. Asterisks denote artifacts.

Figure 2. Diffuse reflectance UVvis spectra of manganese-promoted sulfated zirconia (2.0 wt % Mn) in a 40 mL/min O2 flow in the temperature range of 323773 K. Asterisks denote artifacts.

in the range of 3001000 nm are thus likely to be associated with manganese species. Two bands are located in the visible range at 580 and 680 nm, and their positions did not change as a function of temperature, which is typical of intracenter dd transitions. The intensities of both bands decreased with increasing temperature, although not in a parallel manner, indicating the presence of two different manganese species. The intensity of dd transitions decreases with increasing temperature,42 but this effect alone cannot explain why the band at 680 nm almost completely disappeared. The number of manganese species responsible for this band must have significantly decreased during heating in inert gas. The most intense feature in the spectra was observed in the UV region, with a maximum in the 300400-nm range. This band originates from ligand-to-metal, specifically, oxygen-tomanganese, charge-transfer (CT) transitions. With time on stream and rising temperature, the intensity of the band in the 300400-nm range decreased, and the maximum of the band shifted to higher energy (Figure 1). Specifically, at a temperature of 323 K (at the beginning of the activation treatment), the

ARTICLE

Figure 3. Diffuse reflectance UVvis spectra of MnSZ (0.5 wt % Mn) at low temperatures (283 K, red curve; 128 K, blue curve) and under rough vacuum (1 mbar). The same two spectra are shown twice, on two different scales (left and right y axes).

maximum was located at 325 nm, whereas at 773 K, the maximum was at 300 nm. The observed shift of the maximum was accompanied by a decrease of the overall width of the absorption feature. Charge transfer to manganese at a higher oxidation state (characterized by lower energy and higher intensity) is replaced by charge transfer to manganese at a lower oxidation state (characterized by higher energy and lower intensity). The overall spectral changes, including the disappearance of dd transitions, thus indicate reduction of manganese, most likely to a valence of 2þ, because the dd transitions of Mn2þ are spin-forbidden. This interpretation is consistent with the previous X-ray absorption near-edge structure (XANES) results, which show that the average valence of manganese decreases by about 0.2 to a value of 2.5 ( 0.05 during heating in inert gas.12 The bands at 580 and 680 nm have been ascribed to dd transitions of Mn2þ or Mn3þ.25 The transition of Mn2þ in octahedral cubic surroundings that is lowest in energy, 6A1 f 4 T1, falls into the range of 400600 nm (Table 1), but it is spinforbidden and should not reach an intensity of the same order of magnitude as ligand-to-metal charge-transfer bands. These considerations allow the conclusion that the bands at 580 and 680 nm do not belong to Mn2þ species. The spectral changes observed during heating of MnSZ (2 wt % Mn) in a flow of O2 are shown in Figure 2. The charge-transfer transition at 320 nm and the band at 580 nm increased noticeably in intensity as the temperature was raised, whereas the band at 680 nm increased marginally, if at all. The position of the maximum of the charge-transfer band remained unchanged with increasing temperature and time on stream. As in the case of helium flow, the difference in the intensity change of the bands at 580 and 680 nm with temperature rise also allows one to assume that these two bands belong to different species. On the other hand, when a sample of MnSZ (0.5 wt % Mn) was cooled to below room temperature in a vacuum (Figure 3), the band at ∼590 nm did not change position and increased in intensity. Taking into account the observed increase in the intensity of this band and the impossibility of change of the oxidation degree of the manganese ion under the applied experimental conditions, one can suggest that the band at ∼590 nm (Figure 3) belongs to a 8102

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

manganese ion with a definite oxidation degree and arises from a dd transition.42 Hence, it appears that the nearby band at 580 nm in MnSZ (2 wt % Mn) can be also assigned to a dd transition. The bands at 580 and 680 nm cannot both be assigned to dd transitions of Mn3þ ions,23 because the bands do not behave similarly as a function of temperature, independent of the atmosphere (Figures 13). Most probably these bands also do not arise from two different Mn3þ species or two different Mn4þ species because the ratio of the ionic radii of the impurity center and of the substituted Zr4þ ion in the host crystal mainly affects the structure of the ligand surroundings. The average valence determined by XANES analysis12 also clearly testifies to the presence of Mn2þ, Mn3þ, and Mn4þ ions. We exclude dd transitions of Zr3þ because they fall into the range of 320380 nm.43,44 With Mn2þ dd transitions excluded and oxidation of the initial Mn2þ beyond a valence of 4þ during calcination unlikely, we consider dd transitions of Mn3þ and Mn4þ species. The crystal field is stronger for the Mn4þ ion than for the Mn3þ ion; for example, in the cubic crystal field, the parameter Dq for Mn4þ ions45 significantly exceeds that for Mn3þ ions. Thus, we ascribe the band at 680 nm to Mn3þ species and the band at 580 nm to Mn4þ species.

derivative of the electronnuclear potential energy W(|r  RR|) at the point q = 0   1 X DWðjr  R R jÞ ð4Þ vðrÞ ¼ pffiffiffi n R DR R R R ¼ R 0R where RR represents the coordinates of the ligand surrounding of a manganese ion, R0R represents their equilibrium values in the absence of electrons in the d shell, R numbers the ligands in the first coordination sphere, and n is the total number of these ligands. The energy W(|r  RR|) appears as a term in the Hamiltonian He. 4.2. Crystal Field Model. Our next step is the choice of the model of the crystal field and the electronvibrational interaction for all manganese species. We employed the exchange charge model of the crystal field4651 that takes into account two contributions to the energy of the manganese ions in the crystal field, namely, the contribution arising from the interaction of the electrons in the d shell of the manganese ions with the point charges (pc) of the surrounding ligands and the contribution coming from the overlap of the 3d orbital with the ligand orbitals. The latter is referred to as the contribution of exchange charges (ec). In this model, the parameters Bm l of the crystal field operator X X Vc ¼ Wðj B r  R BR jÞ ¼ Bml Ylm ðϑ, jÞ ð5Þ R

4. MODEL OF MANGANESE IONS IN THE ZIRCONIA LATTICE

are represented as mðpcÞ

Bml ¼ Bl

4.1. Hamiltonian of the System. We employ the quasimo-

lecular (cluster) model that considers a complex formed by a central ion, in this case manganese, and ligands, in this case the adjacent oxygen ions of the lattice. The displacements of the ligands are assumed to be a result of the modulation of the crystal field acting on the manganese ion by lattice vibrations. The total Hamiltonian for a manganese ion is represented as H ¼ He þ Hυ þ Heυ

ð1Þ

where He is the electronic Hamiltonian determining the wave functions and the eigenvalues of a manganese ion in a fixed nuclear configuration; this Hamiltonian includes the kinetic energy of the electrons in the d shell as well as their interaction with the ligand surroundings. Hυ is the Hamiltonian of free lattice vibrations. We account only for the interaction of the manganese ions with the fully symmetric breathing mode of the nearest surroundings because this mode produces the maximum change of the potential energy per unit displacement. In this case, the Hamiltonian Hυ takes the form ! pω 2 D2 q  2 Hυ ¼ ð2Þ 2 Dq where q is the symmetry-adapted vibrational coordinate corresponding to the fully symmetric (breathing) vibration and ω is the frequency of this vibration. Heυ is the vibronic interaction of the manganese ion with the fully symmetric vibration. In our model, this interaction is assumed to be linear with respect to atomic displacements and can be represented in the form Heυ ¼ vðrÞq

l, m

ð3Þ

where r represents the set of electronic coordinates. The function ν(r) (having the dimensions of energy) is determined as the

mðecÞ

þ Bl

where the component    X ZR e2 r l m 4π mðpcÞ ¼ Y ðϑR , jR Þ Bl 2l þ 1 R ðRR Þl þ 1 l

ð6Þ

ð7Þ

is determined in the usual manner4751 and depends on the positions of the ligands, their effective charges, and radial integrals calculated with the radial wave functions specified in ref 52 for the manganese ions. The component Bm(ec) l 2 X 8πe Sl ðRR Þ m mðecÞ G Bl ¼ Y ðϑR , jR Þ ð8Þ RR l 5 R is proportional to a linear combination of squares of different types of overlap integrals of the 3d functions of the manganese ions and 2s and 2p wave functions of the oxygen ligands:52 Sp(RR) = [Ss(RR)]2 þ [Sσ(RR)]2 þ γp[Sπ(RR)]2, with γ2 = 1 and γ4 = 4/3; Ss(RR) = Æ3d,m = 0|2sæ; Sσ(RR) = Æ3d,m = 0|2p,m = 0æ; and Sπ(RR) = Æ3d,m = (1|2p,m = (1æ are overlap integrals of the 3d wave functions of manganese ions and the 2s and 2p wave functions of the oxygen ligands, G is the phenomenological parameter of the model, and Ypm(ϑ,j) represents normalized spherical harmonics. The operator ν(r) of the interaction with the fully symmetric (breathing) mode can be obtained by substitution of the crystal field potential (eq 5) into eq 4 and and Bm(ec) over the differentiation of the parameters Bm(pc) l l radii RR. 4.3. Form Function of the Optical dd Transition in the Semiclassical Approximation. In the subsequent sections, the calculation of the crystal field splittings and vibronic coupling constants determining the profiles of the absorption bands of Mn3þ and Mn4þ are performed in the exchange charge model of the crystal field with the aid of the relations listed above. We 8103

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

employ the adiabatic approximation, neglecting the nuclear kinetic energy in the Hamiltonian Hυ (eq 2). The validity of this approximation for the optical band problem is restricted to the case of strong vibronic coupling or high temperatures.42 Because the optical spectra under examination were measured in the temperature range of 373723 K, this approximation is applicable. In the adiabatic approximation, the energies of the ground states i and the excited states f of the manganese ions represent the adiabatic potentials Ui ðqÞ ¼

pωq2 þ υi Q , 2

Uf ðQ Þ ¼ Δf þ υf Q þ

pωQ 2 ð9Þ 2

where υi = Æi|ν(r)|iæ and υf = Æf|ν(r)|fæ are the matrix elements of the operator of electronvibrational interaction (eq 4) in the ground and excited states, respectively. In the adiabatic approximation, the form function F(Ω) that allows for all intracenter optical transitions from the ground state i to the excited states f looks as follows Z ¥ 1X jÆijudeff jf æj2 FðΩÞ ¼ Z f ¥   Ui ðqÞ exp  ð10Þ δ½Uf ðqÞ  Ui ðqÞ  pΩ dq kB T where deff and u are the dipole moment and polarization vectors, respectively; Ω is the frequency of light; the symbol |...|2 means averaging over light polarization  Z¼

2kB Tπ pω

1=2

2

! !3 2 2 X  Δf  υ υ f i 4exp 5 exp exp  þ 2kB Tω kB T 2kB Tpω f

ð11Þ is the partition function; and Δf is the crystal field gap between the excited (f) and ground (i) state. Carrying out the integration in eq 10 for the form function Fdd(Ω) that allows for intracenter optical transitions between the ground i and excited f states of manganese ions, we obtain X jÆijudeff jf æj2 FðΩÞ ¼ Z1 f

!   υi 2 2kB Tπ 1=2 exp jυf  υi j1 pω 2kB Tpω 2 !2 3 υi ðυi  υf Þ 5 pω exp4 Δf  pΩ þ pω 2kB Tðυi  υf Þ2 ð12Þ The energy of the maximum of the dd band arising from the transition i f f pΩif ¼ Δf þ

υi ðυi  υf Þ pω

ð13Þ

should be attributed to the observed gap between the d levels involved in the optical transition. The gap Δf is redetermined as a result of the d electron shifting the ligands to new equilibrium positions. In the subsequent sections, we calculate the apparent absorption spectra with the aid of the relation K(Ω) ≈ ΩF(Ω). We describe how we evaluate the energy spectrum, the vibronic coupling constants, and the apparent absorption spectra of the Mn3þ and Mn4þ ions in manganese-promoted sulfated zirconia.

5. CRYSTAL FIELD SPLITTINGS AND VIBRONIC INTERACTION OF MN3þ IONS One consequence of the incorporation of promoter ions within the zirconia lattice is the generation of oxygen vacancies for charge compensation (resulting from the lower valence of promoter ions in comparison to Zr4þ). The vacancies provoke atomic displacements of the nearest or next-nearest neighbors, and in consequence, an increase of the impurity concentration leads to a change in the spatial symmetry of the whole doped crystal. Much research has already been done to understand the changes in the structure of zirconia upon doping,53,54 but agreement regarding the magnitude and direction of the displacements of the oxygen and metal ions from their lattice positions is lacking. Such a situation arises because of the variety of compositions examined and different thermal histories of the samples. However, the main reason for the absence of detailed information on crystal rearrangement is the limited ability of experimental methods. Single-crystal studies using integrated intensities of Bragg reflections can provide information only about “the average structure” in which all unit cells are superimposed. Because of this superposition, it is sometimes difficult to detect subtle atomic displacements occurring randomly throughout the crystal. Another experimental difficulty consists of resolving the displacements of zirconium and dopant cations. Therefore, in the present work, complementary methods were applied to provide insight into the structure of promoted zirconia oxide. This study is based on the combination of modeling the dopantvacancy clusters and experimental optical absorption spectra. A literature review shows that, on the basis of the experimental data (X-ray absorption spectroscopy, X-ray and neutron diffraction, electron paramagnetic resonance), it is possible to predict the number of vacancies in the nearest surroundings of the impurity ion, the number of its nearest-neighbor oxygen ions, the directions of the oxygen-ion displacements, and the symmetry of the nearest surroundings of the impurity ion.43,44,55,56 Neutrondiffraction measurements55 indicate that the presence of oxygen vacancies causes oxygen-ion displacements along the Æ100æ direction toward the vacancies (Figure 4). On the basis of data from single-crystal X-ray diffraction using synchrotron radiation and extended X-ray absorption fine structure (EXAFS) obtained for yttria-doped cubic stabilized zirconia, it was proposed that, in the nearest 8-fold surroundings of the Y3þ ion, two oxygen ions positioned along the body diagonal of the cube are missing (Figure 5), so that the Y3þ ion has only six oxygen ligands.56 EPR and optical spectra of Zr3þ in cubic stabilized zirconia were also successfully explained using a model that considers 6-foldcoordinated Zr3þ ions.43,44 Electron energy loss spectroscopy fine structure analysis of the Mn L23 edge provided evidence that, in manganese-stabilized cubic zirconia, manganese is octahedrally coordinated.57 We assume that introduction of Mn3þ ions instead of Zr4þ ions into the zirconia lattice induces oxygen vacancies for charge compensation and that the nearest surroundings of Mn3þ ions in MnSZ consists of six oxygen ions that are shifted toward the two oxygen vacancies (Figure 5). This assumption is in line with the results obtained in the examination of yttriadoped cubic stabilized zirconia.43,44,56 To maintain an overall neutral state of the solid, oxygen vacancies and Mn3þ would have to exist in a ratio of 1:2; however, there is always a significant fraction of Mn2þ, and vacancies in the model might be shared with adjacent metal ions with a valence lower than 8104

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Figure 4. Oxygen ion displacements toward the vacancy along the 100 direction.

Figure 6. Top view of the Mn3þ complex.

Table 2. Spherical Coordinates of the Ligands Surrounding the Mn3þ Ions R (Å)

Figure 5. Structural model of an M3þ metal center in zirconia showing the different displacements η1 and η2 of the two terns of oxygen atoms toward the oxygen vacancies. a is the cube edge.

4þ. In Figure 6, the top view of the Mn3þ complex is given. Following refs 43, 44, and 56, we model the local rearrangements of the oxygen anions as Æ100æ displacements η1 and η2 of the two types of oxygens toward each of the two vacancies within a cube (Figure 5). In this approximation, the spherical angles ϑi of the six oxygens and the distances between the central Mn3þ ion and the two types of oxygen ions are expressed as ffi 1pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 4ηi 2  4aηi þ 3a2 , Ri ¼ 2   a þ 2η ði ¼ 1, 2Þ θi ¼ arccos ð 1Þi þ 1 pffiffiffi i 2 3 Ri

ð14Þ

where a is the cube edge and η1 and η2 are the displacements of the oxygen ions shown in Figure 5. The spherical coordinates of the ligand surrounding of the Mn3þ ions are listed in Table 2. The dopant ions represent centers of small radii, and it is sufficient to account for interaction of these ions with ions of their first coordination sphere.4651 Later, we calculate the operators of the crystal field and electronvibrational interaction in this approximation assuming that the point charge of the oxygen ligands is equal to 2e. For Mn3þ species, the crystal field operator of C3v symmetry written in the frame of reference related to the C3 axis (z axis) shown in Figure 5 is

θi (deg)

1

R2

0

θ2

2

R1

300

θ1

4

R1

60

θ1

6 7

R2 R1

240 180

θ2 θ1

8

R2

120

θ2

1 3þ Table 3. Crystal Field Parameters Bm l (in cm ) for Mn Calculated with η1 = 0.17 Å, η2 = 0.23 Å, and G = 3.65

pc B02 B04 B34

ec

total

21090

8255

29345

1282

3873

5155

4760

14330

19090

where Bm l represents the parameters of the crystal field. The numerical values of these parameters estimated in the exchangecharge model of the crystal field with the aid of eqs 68 are reported in Table 3, which also identifies the partial contributions of the exchange and point charges. In calculations of the crystal field, the values G = 3.65, η1 = 0.17 Å, and η2 = 0.23 Å were taken for the phenomenological parameters of the model, G (eq 8) and ηi (i = 1, 2) (eq 14). These values were obtained from comparison of the calculated and observed band maxima. As an illustration, the analytical expres3þ complex are given in sions for the parameters Bm l for the Mn Appendix 1 (eqs A.1A.3). In the trigonal crystal field, the ground state of the Mn3þ ion is a doublet 5E(1) (Figure 7), whereas the excited cubic term of this ion 5T2 is split into two levels 5A1 and 5E(2). The wave functions of these levels have the form j5 A 1 æ ¼ jx1 x1 u1 u1 j, ð1Þ

ð1Þ

j5 Eð1Þ u æ ¼ C11 jx1 x0 x1 u1 j þ C12 jx1 x0 u1 u1 j, ð1Þ

ð1Þ

j5 Eð1Þ uþ æ ¼ C12 jx1 x0 x1 u1 j  C11 jx1 x0 u1 u1 j,

Vcr ¼ B02 Y2, 0 ðθ, jÞ þ B04 Y4, 0 ðθ, jÞ þ B34 ½Y4, 3 ðθ, jÞ  Y4, 3 ðθ, jÞ

ji (deg)

ð2Þ

ð2Þ

j5 Eð2Þ u æ ¼ C11 jx1 x0 x1 u1 j þ C12 jx1 x0 u1 u1 j, ð2Þ

ð2Þ

j5 Eð2Þ uþ æ ¼ C12 jx1 x0 x1 u1 j  C11 jx1 x0 u1 u1 j

ð15Þ 8105

ð16Þ

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Table 4. Vibronic Coupling Constants (in cm1) for Mn3þ Calculated with η1 = 0.17 Å, η2 = 0.23 Å, and G = 3.65 pc

Figure 7. Crystal field splittings of the Mn3þ ion calculated with the parameters η1 = 0.17 Å, η2 = 0.23 Å, and G = 3.65.

ec

total

υ( A1)

420

1650

1230

υ(5E(2))

535

1045

510

υ(5E(1))

740

1870

1130

5

The numerical values of the coefficients of the wave functions (1) (2) (eq 16), namely, C(1) 11 = 0.9881, C12 = 0.1536, C11 = 0.1536, (2) and C12 = 0.9881, were obtained by diagonalization of the crystal field potential (eq 15). The trigonal one-electron states (complex trigonal basis) relevant to the trigonal coordinate system (Figures 5 and 6) are expressed in terms of the d functions ξ = yz, η = xz, ζ = xy(t2g), u = 3z2  r2, υ = (3)1/2(x2  y2)(eg) defined in the tetragonal basis as follows 8   1 > > > jxþ æ ¼  pffiffiffi ðεξ þ εη þ ζÞ, > > > ( pffiffiffi  3 > < 1 uþ ¼  ð1= 2Þðu þ iυÞ, pffiffiffi jx æ ¼ pffiffiffi ðεξ þ εη þ ζÞ, ε ¼ expð2πi=3Þ u ¼ ð1= 2Þðu  iυÞ, > >  3 > > > 1 > > : jx0 æ ¼ pffiffiffi ðξ þ η þ ζÞ, 3

ð17Þ The energy gaps ΔA = E(5A1)  E(5E(1)) = 13880 cm1 and ΔE = E(5E(2))  E(5E(1)) = 11420 cm1 were calculated using the crystal field potential (eq 15), the wave functions (eq 16), and the crystal field parameters given in Table 3. Thus, the absorption band originating from the Mn3þ species can be assigned to the optical transitions 5E(1) f 5E(2) and 5E(1) f 5A1. To obtain the scheme of the adiabatic potential sheets and to calculate the shape of the absorption band arising from the Mn3þ species with the aid of eqs 58, we estimated the corresponding vibronic coupling constants υ(5A1), υ(5E(2)), and υ(5E(1)). In Table 4, the vibronic constants for various states of the Mn3þ ion are given and broken down into the contributions from the electrostatic field of point ligand charges and the exchange charge field. It can be seen that the main contribution to the vibronic coupling constants comes from the exchange charge field. The positions of the dd bands that form the total absorption spectrum of the Mn3þ ions can be illustrated by the scheme of the adiabatic potentials as shown in Figure 8, where q is the fully symmetric coordinate (the breathing mode) of the local surrounding of the Mn3þ ions in sulfated zirconia. The pattern of the adiabatic potentials pω UEð1Þ ðqÞ ¼ υð E Þq þ q2 , 2 5 ð1Þ

pω 2 q, 2 pω UA ðqÞ ¼ ΔA þ υð5 A 1 Þq þ q2 2

Figure 8. Adiabatic potentials for the Mn3þ ions calculated with the parameters η1 = 0.17 Å, η2 = 0.23 Å, G = 3.65, and pω = 615 cm1.58 The wavelengths λi indicate the maxima of the absorption bands arising on the transitions from the ground adiabatic potential sheet to the excited ones.

6. OSCILLATOR STRENGTHS FOR THE TRANSITIONS OF MN3þ IONS To calculate the shape of the absorption band arising from the Mn3þ ions, we consider the intensities of the optical transitions of these ions. The states of the Mn3þ ion belong to even parity. The parity-forbidden electric-dipole transitions between these states become allowed because of the presence of the static oddparity potential Vodd for the Mn3þ complexes in sulfated zirconia that have no inversion center. The effective dipole moment deff can be presented as45 deff 

2Vodd d Δ

ð19Þ

where Δ is the effective energy gap between the 3d and oddparity levels in the energy spectrum; d is the electric dipole moment, expressed as d ¼  d k^þ  dþ k^ þ d0 k^0 pffiffiffiffiffiffi er 4π dþ ¼  pffiffiffi Y1, 1 ðθ, jÞ, 3

UEð2Þ ðqÞ ¼ ΔE þ υð5 Eð2Þ Þq þ

ð18Þ

depends on the energy gaps ΔA and ΔE between the electronic states in the crystal field and the vibronic coupling constants. In Figure 8, the maxima of the absorption bands arising from the transitions from the ground adiabatic potential sheet to the excited ones are indicated. It is seen that Mn3þ species contribute to the visible and NIR parts of the spectra.

pffiffiffiffiffiffi er 4π d ¼  pffiffiffi Y1, 1 ðθ, jÞ, 3 pffiffiffiffiffiffi er 4π ð20Þ d0 ¼  pffiffiffi Y1, 0 ðθ, jÞ 3 √ √ ^ Finally, and k^þ = (1/ 2)(i þ ij), k^ = (1/ 2)(i  ij), k^0 = k. i, j, and k are the unit vectors in the directions x, y, and z, respectively, of the trigonal coordinate axes (Figures 5 and 6). To estimate the oscillator strengths of the transitions 5E(1) f 5E(2) and 5E(1) f 5A1 with the aid of eq 19, we use the odd-parity crystal field potential Vodd that corresponds to the trigonal 8106

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Table 5. Distances between the Mn4þ Ion and the NearestNeighbor Oxygen Ions and Spherical Angles of the Oxygen Ions58

Figure 9. Geometrical structure of the Mn4þ center: large shaded circles, Mn4þ; small shaded circles, oxygen.

symmetry of the MnO6 complex. The general form of this operator in a first approximation is Vodd ¼ aðrÞ Y1, 0 ðθ, jÞ

ð21Þ

where a(r) represents the radial part of the potential Vodd. In fact, in the point-charge model of the crystal field, this term represents the main contribution to the oscillator strengths. Using eqs 1921, the matrix elements of the effective dipole moment for all optical transitions in Mn3þ ions were calculated. Omitting the details of the calculations, we quote only the normalized values , X 2 ð1Þ 0 ¼ jÆΓγ jdeff jE γæj f5 ð1Þ E

2 4

X γ, γ0



γ, γ0

5 ð1Þ jÆ5 Eð2Þ γ0 jdeff j Eγ æj þ 2

X γ

3 jÆ5 A 1 jdeff j5 Eð1Þ γ æj

25

ð22Þ

of the matrix elements of the dipole moment for different optical transitions in Mn3þ species: f5E(1) f 5E(2) = 0.97 and f5E(1) 5 f A 1 = 0.03. It is worth noting that the oscillator strength of the 5E(1) f 5A1 transition is small compared with that of the 5 (1) E f 5E(2) transition. The simulation of the absorption coefficient was performed using the relation KMn3þ(Ω) ≈ xMn3þΩFMn3þ(Ω) (where xMn3þ is the relative concentration of Mn3þ ions) and eqs 11 and 12. In the calculations, the parameters given in Table 4, the values f5E(1) f Γ, and the energy splittings shown in Figure 7 were employed. The total calculated and experimental spectra arising from Mn3þ and Mn4þ species are discussed in section 8.

7. CRYSTAL FIELD SPLITTINGS, VIBRONIC INTERACTION, AND INTENSITIES OF OPTICAL TRANSITIONS OF MN4þ IONS When Mn4þ substitutes for Zr4þ, the number of oxygen ligands surrounding the metal cation is assumed to remain unchanged at eight (Figure 9). Modeling of the O2 ions surrounding the Mn4þ ions was performed in the following way: (i) The distances between eight oxygen ions and the Zr4þ ion in tetragonal stabilized zirconia were calculated from the structural data in ref 58. These radii take on the following values: O1, R1 = 2.366 Å; O2, R2 = 2.083 Å; O3, R3 = 2.366 Å; O4, R4 = 2.083 Å; O5, R5 = 2.083 Å; O6, R6 = 2.366 Å; O7, R7 = 2.083 Å; O8, R8 = 2.366 Å. (ii) The radius of the Zr4þ ion is equal to RZr4þ

Ri (Å)

ji (deg)

θi (deg)

1

2.04

315

120

2

2.04

225

120

3

2.04

135

141

4

2.04

45

120

5 6

1.80 1.80

315 225

50 50

7

1.80

135

71

8

1.80

45

50

= 0.82 Å, and the radius of the O2 ion is 1.36 Å. Thus, the mean distance between these ions is 2.18 Å. The radius of the Mn4þ ion is RMn4þ = 0.52 Å, resulting in a mean Mn4þO2 distance of 1.88 Å. (iii) Assuming that the substitution of Zr4þ by Mn4þ changes only the distances between the metal ion and the oxygen ligands, the following approximate relations, where the radii are given in Å, were used to determine the MnO distances in the complex Mn4þO8 1:88=2:18 ¼ x=2:366 1:88=2:18 ¼ y=2:083

ð23Þ

The resulting numerical values for the eight MnO bonds Ri are given in Table 5. The values of the spherical angles θi and ji for the oxygen ligands (Table 5) were taken to be the same as for the complex ZrIVO8.58 The crystal field potential for the Mn4þ ions is Vcr ðθ, jÞ ¼ B02 Y2, 0 ðθ, jÞ þ B04 Y4, 0 ðθ, jÞ þ B12 ½Y2, 1 ðθ, jÞ  Y2, 1 ðθ, jÞ þ iC12 ½Y2, 1 ðθ, jÞ þ Y2, 1 ðθ, jÞ þ iC22 ½Y2, 2 ðθ, jÞ  Y2, 2 ðθ, jÞ þ B14 ½Y4, 1 ðθ, jÞ  Y4, 1 ðθ, jÞ þ iC14 ½Y4, 1 ðθ, jÞ þ Y4, 1 ðθ, jÞ þ iC24 ½Y4, 2 ðθ, jÞ  Y4, 2 ðθ, jÞ þ B34 ½Y4, 3 ðθ, jÞ  Y4, 3 ðθ, jÞ þ iC34 ½Y4, 3 ðθ, jÞ þ Y4, 3 ðθ, jÞ þ B44 ½Y4, 4 ðθ, jÞ þ Y4, 4 ðθ, jÞ

ð24Þ

As in section 5, the parameters of the crystal field potential for the Mn4þ ion were estimated in the exchange-charge model of the crystal field using eqs 68. The phenomenological parameter G = 6.3 that enters the exchange-charge part of the crystal field (eq 8) was determined from comparison of the calculated and observed maxima of the band arising from Mn4þ species. The m numerical values of the parameters Bm l and Cl and the contributions of the exchange and point charges to these parameters are given in Table 6. The crystal field potential (eq 24) mixes the ground 4T1 (t22e), first excited 4T2 (t22e), and second excited 4A2 (t32) cubic terms of the Mn4þ ion and splits them into seven nondegenerate levels with wave functions ji ¼ 8107

7 X k¼1

ðiÞ

Ck Ψ k

ð25Þ

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Table 6. Crystal Field Parameters (in cm1) for the Mn4þ Species B02

B04

B12

B14

B34

B44

pc

2265

4750

3760

1345

160

4300

ec total

3670 5935

16580 21330

2185 5945

4620 5965

400 560

15980 20280

C24

C34 160

C12

C22

C14

pc

3760

450

1345

480

ec

2185

220

4620

1675

400

total

5945

670

5965

2155

560

Table 7. Numerical Values of the Expansion Coefficients C(i) k of the Wave Functions of the Mn4þ Ion ji

C(i) 1

C(i) 2

C(i) 3

0.0286

0.0286

0

j2 0.1244 0.0271 0.0271

0

j1 0.0062

0.0220 0.0220

C(i) 4

C(i) 5

C(i) 6

C(i) 7

0.0862

0.0862

0.9917

0.6959

0.6959 0.1202

0.0921 0.7038 0.7038

j3

0

j4 j5

0 0.0842 0.0842 0.9890 0.0621 0.0621 0.2605 0.6821 0.6821 0 0.0002 0.0002

j6

0

0.7017

0.7017 0.1158 0.0295 0.0295

j7 0.9574 0.1823 0.1823

0

0.0909 0.0909

0 0 0.0409 0 0.0203

Figure 11. Adiabatic potentials of the Mn4þ species in the single-mode model, calculated with G = 6.3, B = 1070 cm1, and pω1 = 665 cm1. The wavelengths λi indicate the maxima of the absorption bands arising on the transitions from the ground adiabatic potential sheet to the excited ones.

Table 8. Vibronic Coupling Constants υi = Æi|V^ A1|iæ (in cm1) for Mn4þ Species pc

ec

total

υ1

90

215

305

υ2 υ3

420 500

1055 1225

1475 1725

υ4

555

790

235

υ5

390

1470

1860

υ6

1430

115

1315

υ7

175

5320

5145

Table 9. Normalized Oscillator Strengths f1fi for the transitions |1æ f |iæ (i = 27) in Mn4þ Species

Figure 10. Crystal field levels of the Mn4þ ion in sulfated zirconia.

where Ψ1 ¼ jt32 , 4 A 2 , S ¼ 3=2, MS ¼ 3=2æ ¼  jξηζj pffiffiffi Ψ2 ¼ jt22 e, 4 T 2ξ , S ¼ 3=2, MS ¼ 3=2æ ¼ ð 3jηζuj þ jηζυjÞ=2 pffiffiffi Ψ3 ¼ jt22 e, 4 T 2η , S ¼ 3=2, MS ¼ 3=2æ ¼ ð  3jζξuj þ jηξυjÞ=2

Ψ7 ¼

¼ 3=2, MS ¼ 3=2æ ¼ jξηuj

ð26Þ

S is the total spin, MS is the spin projection, and the conventional notation |...| is used for the Slater determinants. The coefficients C(i) k of the wave functions (eq 25) are given in Table 7, and the

f

transition

1f2 1f3

0.11 0.16

1f5 1f6

1f4

0.01

1f7

f 0.32 0.40 ∼0

Table 10. Relative Concentrations xMn2þ, xMn3þ, and xMn4þ during Activation of MnSZ (2.0% Mn) in Helium Determined Using UVVis Spectroscopy (Figure 1); Average Valence, R, as Determined by XANES Analysis12; and Calculated Apparent Absorption Spectra in the Visible Range

Ψ4 ¼ jt22 e, 4 T 2ζ , S ¼ 3=2, MS ¼ 3=2æ ¼  jξηυj pffiffiffi Ψ5 ¼ jt22 e, 4 T 1R , S ¼ 3=2, MS ¼ 3=2æ ¼ ð 3jηζυj  jηζujÞ=2 pffiffiffi Ψ6 ¼ jt22 e, 4 T 1β , S ¼ 3=2, MS ¼ 3=2æ ¼  ð 3jζξυj þ jζξujÞ=2 jt22 e, 4 T 1γ , S

transition

T (K)

β

373

1.3

523 623

1.35 1.4

723

1.54

R

xMn2þ

xMn3þ

xMn4þ

2.72

0.54

0.20

0.26

2.63 2.57

0.60 0.64

0.17 0.15

0.23 0.21

2.53 (703 K)

0.67

0.13

0.20

crystal field levels of the Mn4þ ions and their energies are illustrated in Figure 10. The numbering of levels in Figure 10 corresponds to that of the wave functions in Table 7. During diagonalization of the crystal field matrix in the basis of states 8108

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

(eq 26), the energy difference ∼12B (where B is the Racah parameter) between the cubic states 4A2(t32) and 4T1(t22e), 4T2(t22e) provided by the Coulomb interaction45 was also taken into account. The shapes of the adiabatic potential sheets for Mn4þ ions (Figure 11) can be easily calculated with the aid of an equation similar to eq 18 and the numerical values of the vibronic coupling constants, given in Table 8 for the states of the Mn4þ ions participating in optical transitions. As for the Mn3þ ion, the vibronic coupling constants for the Mn4þ ion were calculated using eqs 48. For each Mn4þ level participating in optical transitions, the vibronic coupling constant is represented as a Table 11. xMn4þ/xMn3þ Ratios during Activation of MnSZ (2% Mn) in O2 Determined Using Diffuse Reflectance UVVis Spectra (Figure 2) and Calculated Apparent Absorption Spectra in the Visible Range T (K)

β = xMn4þ/xMn3þ

373

1.33

523

1.47

623

1.5

723

1.6

sum of contributions arising from the point-charge and exchange-charge crystal fields (Table 8). The pattern of adiabatic potentials (Figure 11) shows that the optical transitions of the Mn4þ ions can take place in a wide spectral range from near-infrared to violet. At the same time, the estimation of the normalized oscillator strengths f1fi = |Æi|V^ (1) oddd| 2 1æ|2/∑k|Æk|V^ (1) d|1æ| using eq 18 and the following expression odd for the odd-parity crystal field potential (written in the same approximation as eq 21) Mn Vodd



¼ A11 Y1, 1 ðθ, jÞ þ A1, 0 Y1, 0 ðθ, jÞ þ A1, 1 Y1, 1 ðθ, jÞ ð27Þ

shows that the optical transitions from the ground adiabatic potential sheet to sheets 5 and 6 are most pronounced (Table 9, Figure 11). Thus, the apparent absorption band arising from Mn4þ species is formed mainly by the optical transitions 1 f 5 and 1 f 6. The absorption band arising from Mn4þ species was calculated using the relation KMn4þ(Ω) ≈ xMn4þΩFMn4þ(Ω), where xMn4þ is the relative concentration of Mn4þ ions and the form function FMn4þ(Ω) is determined by eq 12.

Figure 12. Apparent absorption spectra of MnSZ (2 wt % Mn) in a flow of helium (normalized). Circles, experimental data; black line, total calculated spectra, dashed line, contribution from the Mn3þ species to the total band, dot-dashed line, contribution from the Mn4þ species to the total band. (a) T = 373 K, xMn4þ/xMn3þ = 1.30; (b) T = 523 K, xMn4þ/xMn3þ = 1.35; (c) T = 623 K, xMn4þ/xMn3þ = 1.40; (d) T = 723 K, xMn4þ/xMn3þ = 1.54. 8109

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

Figure 13. Apparent absorption spectra of the MnSZ (2 wt % Mn) in flow of O2 (normalized). Circles, experimental data; black line, total calculated spectra; dashed line, contribution from the Mn3þ species to the total band; dot-dashed line, contribution from the Mn4þ species to the total band. (a) T = 373 K, xMn4þ/xMn3þ = 1.33; (b) T = 523 K, xMn4þ/xMn3þ = 1.47; (c) T = 623 K, xMn4þ/xMn3þ = 1.50; (d) T = 723 K, xMn4þ/xMn3þ = 1.60.

8. DETERMINATION OF THE RELATIVE CONCENTRATIONS OF MN2þ, MN3þ, AND MN4þ IONS IN MANGANESE-PROMOTED SULFATED ZIRCONIA Following the considerations presented in section 2, we simulated the apparent absorption spectra of manganese-promoted sulfated zirconia in the wavelength range of 450 nm < λ < 780 nm using the relation K(Ω) ≈ xMn3þΩFMn3þ(Ω) þ xMn4þΩFMn4þ(Ω). The simulation consisted of two steps. In the first step, we calculated the form functions FMn3þ(Ω) and FMn4þ(Ω) using eqs 11 and 12 and data listed in the tables. In the second step, we determined the ratio xMn4þ/xMn3þ for a variety of conditions by fitting calculated to experimental spectra. Such fits are shown in Figure 12 for spectra obtained during activation of manganese-promoted sulfated zirconia in helium flow. At 373 K, the xMn4þ/xMn3þ ratio was 1.30; at 523 K, it was 1.35; and at 723 K, it reached 1.54 (Table 10). In flowing O2 (Figure 13), the same trend was observed: in the temperature range of 373723 K the xMn4þ/xMn3þ ratio increased from 1.33 to 1.60 (Table 11). The xMn4þ/xMn3þ values obtained for different temperatures were further used to determine the relative concentrations xMn2þ, xMn3þ, and xMn4þ of Mn2þ, Mn3þ, and Mn4þ ions, respectively, in the catalyst at various temperatures. For this purpose, the

following system of three equations was solved 2xMn2þ þ 3xMn3þ þ 4xMn4þ ¼ R, xMn2þ þ xMn3þ þ xMn4þ ¼ 1, xMn4þ =xMn3þ ¼ β

ð28Þ

The first equation determines the mean valence R of manganese ions in the catalyst. The second equation represents the normalization condition, and the third equation characterizes the ratio of xMn3þ and xMn4þ ions. To determine the values xMn2þ, xMn3þ, and xMn4þ from the system of eqs 28, the mean valence R at various temperatures was taken from our preceding work,12 in which equivalently prepared samples were analyzed by X-ray absorption spectroscopy (XAS) under conditions similar to those applied here. The value R varied with the experimental conditions. The determined values xMn2þ, xMn3þ, and xMn4þ are given in Table 10. From the data in Table 10, it follows that the relative concentration of Mn2þ ions increases from 0.54 to 0.67 during heating in helium flow, which is in accordance with the temperature behavior of the charge-transfer band. The edge of the charge-transfer band shifts to shorter wavelengths (or higher energy), indicating an increase in the number of Mn2þ ions. 8110

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A

ARTICLE

The fraction of Mn3þ ions decreased from 0.20 to 0.13, which is also in agreement with the CT band behavior. Mn3þ species seem more prone to reduction than Mn4þ species. The XAS data indicate that, during heating in 50% O2, a net reduction occurs.12 No XAS data and thus no average valence were available for heating in 100% O2. However, the data in Table 11 indicate that the relative concentration of Mn4þ increases, and the spectra in Figure 2 demonstrate also that the absolute concentration of Mn4þ increases whereas that of Mn3þ remains constant, indicating that, in 100% O2, a net oxidation occurs.

9. CONCLUSIONS UVvis spectra of manganese-doped zirconia were interpreted. Bands at 300320 nm were assigned to charge transfer from O2 to Mn2þ, Mn3þ, and Mn4þ ions, with the wavelength becoming shorter as the manganese valence decreased. The band at 580 nm mainly arises from the 1f5 and 1f6 transitions in 8-coordinated Mn4þ ions, whereas the band at 680 nm was assigned to Mn3þ species in trigonally distorted octahedral symmetry. These findings make it possible to re-evaluate contradictory interpretations in the literature. The crystal field splitting, vibronic interaction, and oscillator strength of Mn3þ and Mn4þ ions in their respective environments of O2 ions were modeled, and the experimental spectra could be well-described using a single phenomenological parameter. With the help of the model, the ratio of Mn3þ to Mn4þ ions in manganese-doped zirconia could be determined from the UVvis spectra, and with additional knowledge of the average valence from XAS data, the fractions of Mn2þ, Mn3þ, and Mn4þ ions could be calculated. Mn3þ is the most redox-active valence inside the zirconia lattice: it is more easily reduced than Mn4þ, and Mn2þ is more easily oxidized to Mn4þ than to Mn3þ. In the present contribution, we have considered manganese ions located essentially in the bulk. The possibility for the Mn3þ and Mn4þ ions to be located on the surface was not examined. However, under the assumption that the structure of the complex consisting of the manganese ion and its nearest surroundings remains unchanged on the surface, the obtained results can be considered reasonable for this case as well. In the more general case, this question requires special consideration. Such consideration will be reported elsewhere. ’ APPENDIX 1 Crystal field parameters for the Mn3þ complex of C3v symmetry " # pffiffiffiffiffiffi 2 X ZÆr 2 æ 5π 24e Gaða  4η ÞS ðR Þ 2 i i B02 ¼ ða2 þ 4aηi Þ  5 16Ri5 8Ri3 i ¼ 1, 2

ðA:1Þ pffiffiffi 8e2 π ¼ 45 " X 5ZÆr 4 æ 4 3 4 3 2 2 9 ð56ηi þ a  136a ηi þ 240a ηi þ 32aηi Þ 64R i i ¼ 1, 2  4 9Gða  136a3 ηi þ 240a2 ηi 2 þ 32aη3i  56ηi 4 ÞS4 ðRi Þ þ 32Ri5 B04

ðA:2Þ

B34 ¼

pffiffiffiffiffiffiffiffi 32e2 70π 45

" X 5ZÆr 4 æ 4 3 4 3 2 2 9 ð2ηi þ a  a ηi  3a ηi þ 5aηi Þ 64R i i ¼ 1, 2  9G4 ða  ηi Þ3 ða þ 2ηi ÞS4 ðRi Þ þ ðA:3Þ 32Ri5

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected] (F.C.J.), [email protected] (S.I.K.). Present Addresses †

Permanent address: Institute of Applied Physics of the Academy of Sciences of Moldova, Academy Street 5, 2028 Kishinev, Moldova. ‡ New address: Chemical, Biological, and Materials Engineering, University of Oklahoma, 100 East Boyd Street, Norman, OK 73019-1004, USA.

’ ACKNOWLEDGMENT Financial support of S.I.K. through DFG Grant 436 MOL 17/ 1/04 is acknowledged. The authors thank the Max Planck Society for long-term financial support of the project and Giulio Lolli for recording the low-temperature UVvis spectra. ’ REFERENCES (1) Lamb, A. B.; Bray, W. C.; Frazer, J. C. W. Ind. Eng. Chem. 1920, 12, 213–221. (2) Kr€amer, M.; Schmidt, T.; St€owe, K.; Maier, W. F. Appl. Catal. A 2006, 302, 257–263. (3) Nozaki, F.; Sodesawa, T.; Yamamoto, T. J. Catal. 1983, 84, 267–9. (4) W€ollner, A.; Lange, F.; Schmelz, H.; Kn€ ozinger, H. Appl. Catal. A 1993, 94, 181–203. (5) Keshavaraja, A.; Ramaswamy, A. V. Appl. Catal. B 1996, 8, L1–L7. (6) Keshavaraja, A.; Ramaswamy, A. V. Chem. Commun. 1996, 397–398. (7) Hsu, C.-Y.; Heimbuch, C. R.; Armes, C. T.; Gates, B. C. J. Chem. Soc., Chem. Commun. 1992, 1645–1646. (8) Lange, F. C.; Cheung, T.-K.; Gates, B. C. Catal. Lett. 1996, 41, 95–99. (9) Lin, C.-H.; Hsu, C.-Y. J. Chem. Soc., Chem. Commun. 1992, 1479–1480. (10) Wan, K. T.; Khouw, C. B.; Davis, M. E. J. Catal. 1996, 158, 311–326. (11) Song, S. X.; Kydd, R. A. Catal. Lett. 1998, 51, 95–100. (12) Jentoft, R. E.; Hahn, A. H. P.; Jentoft, F. C.; Ressler, T. Phys. Chem. Chem. Phys. 2005, 7, 2830–2838. (13) Klose, B. S.; Jentoft, F. C.; Schl€ogl, R.; Subbotina, I. R.; Kazansky, V. B. Langmuir 2005, 21, 10564. (14) Klose, B. S.; Jentoft, F. C.; Schl€ogl, R. J. Catal. 2005, 233, 68. (15) Li, X.; Nagaoka, K.; Simon, L. J.; Olindo, R.; Lercher, J. A.; Hofmann, A.; Sauer, J. J. Am. Chem. Soc. 2005, 127, 16159. (16) Valigi, M.; Gazzoli, D.; Dragone, R.; Marucci, A.; Mattei, G. J. Mater. Chem. 1996, 6, 403–408. (17) Ressler, T.; Brock, S. L.; Wong, J.; Suib, S. L. J. Phys. Chem. 1999, 193, 6407–6420. 8111

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112

The Journal of Physical Chemistry A (18) Jentoft, F. C.; Hahn, A.; Kr€ohnert, J.; Lorenz, G.; Jentoft, R. E.; Ressler, T.; Wild, U.; Schl€ogl, R.; H€assner, C.; K€ohler, K. J. Catal. 2004, 224, 124–137. (19) Goldberg, D. P.; Telser, J.; Krzystek, J.; Garrido Montalban, A.; Brunel, L.-C.; Barrett, A. G. M.; Hoffman, B. M. J. Am. Chem. Soc. 1997, 119, 8722. (20) Wallbank, B.; Main, I. G.; Johnson, C. E. J. Electron Spectrosc. Relat. Phenom. 1974, 5, 259. (21) Oku, M.; Hirokawa, K.; Ikeda, S. J. Electron Spectrosc. Relat. Phenom. 1975, 7, 465. (22) Pratt, G. W., Jr.; Coelho, R. Phys. Rev. 1959, 2, 281. (23) Fernandez Lopez, E.; Sanchez Escribano, V.; Resini, C.; Gallardo-Amores, J. M.; Busca, G. Appl. Catal. B: Environ. 2001, 29, 251. (24) Milella, F.; Gallardo-Amores, J. M.; Baldi, M.; Busca, G. J. Mater. Chem. 1998, 8, 2525. (25) Sjoerd Kijlstra, W.; Poels, E. K.; Bliek, A.; Weckhuysen, B. M.; Schoonheydt, R. A. J. Phys. Chem. B 1997, 101, 309. (26) Mehra, A.; Venkateswarlu, P. J. Chem. Phys. 1968, 48, 4381. (27) Geschwind, S.; Kisliuk, P.; Klein, M. P.; Remeika, J. P.; Wood, D. L. Phys. Rev. 1962, 126, 1684. (28) Holmes, O. G.; McClure, D. S. J. Chem. Phys. 1957, 26, 1686. (29) McClure, D. S. J. Chem. Phys. 1962, 36, 2757. (30) Rajic, N.; Stojakovic., D.; Hocevar, S.; Kaucic, V. Zeolites 1993, 13, 384–387. (31) Cotton, F. A.; Wilkinson, G.; Murillo, C. A.; Bochmann, M. Advanced Inorganic Chemistry, 6th ed.; Wiley-Interscience: New York, 1999. (32) Cordischi, D.; Valigi, M.; Gazzoli, D.; Indovina, V. J. Solid State Chem. 1975, 15, 82–88. (33) St€ocker, H. J. Ann. Chim. 1960, 5, 1459–1502. (34) St€ocker, J.; Collongues, R. C. R. Seances Acad. Sci. 1957, 245, 695–697. (35) Keshavaraja, A.; Ramaswamy, A. V. J. Mater. Res. 1994, 9, 837–840. (36) Jentoft, F. C.; Hahn, A.; Kr€ohnert, J.; Lorenz, G.; Jentoft, R. E.; Ressler, T.; Wild, U.; Schl€ogl, R.; H€assner, C.; K€ohler, K. J. Catal. 2004, 224, 124–137. (37) Hahn, A.; Ressler, T.; Jentoft, R. E.; Jentoft, F. C. Chem. Commun. 2001, 537–538. (38) Ciuparu, D.; Ensuque, A.; Shafeev, G.; Bozon-Verduraz, F. J. Mater. Sci. Lett. 2000, 19, 931. (39) Navio, J. A.; Hidalgo, M. C.; Colon, G.; Botta, S. G.; Litter, M. I. Langmuir 2001, 17, 202. (40) Gutierrez-Alejandre, A.; Ramirez, J.; Busca, G. Catal. Lett. 1998, 56, 29. (41) Hintze, P. E.; Kjaergaard, H. G.; Vaida, V.; Burkholder, J. B. J. Phys. Chem. A 2003, 1112–1118. (42) Perlin, Yu. E.; Tsukerblat, B. S. Effects of Vibronic Interactions in the Optical Spectra of Paramagnetic Impurity Ions; Shtiintsa: Kishinev, USSR, 1974. Perlin, Yu. E.; Tsukerblat, B. S. Optical Bands and Polarization of JahnTeller Centers. In The Dynamical JahnTeller Effect in Localized Systems; Perlin, Yu. E., Wagner, M., Eds.; Elsevier: New York, 1984; Vol. 7, p 251. (43) Azzoni, C. B.; Paleari, A. Phys. Rev. B 1991, 44, 6858. (44) Azzoni, C. B.; Bolis, L.; Paleari, A.; Samoggia, G.; Scardina, F. Phys. Rev. B 1995, 51, 15942. (45) Sugano, S.; Tanabe, Y.; Kamimura, H. Multiplets of Transition Metal Ions in Crystals; Academic Press: New York, 1970. (46) Popova, M. N.; Chukalina, E. P.; Malkin, B. Z.; Saikin, S. K. Phys. Rev. B 2000, 61, 7421. (47) Klokishner, S.; Melsheimer, J.; Jentoft, F. C.; Schl€ogl, R. Phys. Chem. Chem. Phys. 2004, 6, 2066. (48) Klokishner, S. I.; Tsukerblat, B. S.; Reu, O. S.; Palii, A. V.; Ostrovsky, S. M. Chem. Phys. 2005, 316, 83. (49) Klokishner, S. I.; Tsukerblat, B. S.; Reu, O. S.; Palii, A. V.; Ostrovsky, S. M. Opt. Mater. 2005, 27, 1445. (50) Klokishner, S. I.; Reu, O. S.; Ostrovsky, S. M.; Palii, A. V.; Kulyuk, L. L.; Tsukerblat, B. S.; Towe, E. J. Mol. Struct. 2007, 838, 133. (51) Klokishner, S.; Reu, O.; Ostrovsky, S.; Palii, A.; Towe, E. J. Phys.: Condens. Matter 2007, 19, 486213.

ARTICLE

(52) Clementi, E.; Roetti, C. At. Data Nucl. Data Tables 1974, 14, 177. (53) Li, P.; Chen, I-Wei; Penner-Hahn, J. E. Phys. Rev. B 1993, 48, 10063. 1993, 48, 10074. 1993, 48, 10082. (54) Stapper, G.; Bernasconi, M.; Nicoloso, N.; Parrinello, M. Phys. Rev. B 1999, 59, 797. (55) Steele, D.; Fender, B. E. F. J. Phys. C 1974, 7, 1. (56) Ishizawa, N.; Matsushima, Y. Acta Crystallogr. B 1999, 55, 726. (57) Dravid, V. P.; Ravikumar, V.; Notis, M. R.; Lyman, C. E. J. Am. Ceram. Soc. 1994, 77, 2758. (58) Lutterotti, L.; Scardi, P. J. Appl. Crystallogr. 1990, 23, 246–252.

8112

dx.doi.org/10.1021/jp2034054 |J. Phys. Chem. A 2011, 115, 8100–8112